Lehrstuhl für Informatik 4 Kommunikation und verteilte Systeme Chapter 2: Basics Chapter 3: Multimedia Systems – Communication Aspects and Services Chapter 4: Multimedia Systems – Storage Aspects • Optical Storage Media • Multimedia File Systems • Multimedia Database Systems Chapter 5: Multimedia Usage Chapter 4.3: Multimedia Database Systems 4.3: Multimedia Database Systems • Multimedia Database Management System • Data Structure • Operations on Data • Integration in a Database Model Page 1 Lehrstuhl für Informatik 4 Kommunikation und verteilte Systeme Database Systems User Application Application User Database Management System (DBMS) Database Data Data Data Data Data Data Chapter 4.3: Multimedia Database Systems Page 2 Lehrstuhl für Informatik 4 Kommunikation und verteilte Systeme Multimedia Database Management System Multimedia System • Main task of a Database Management System (DBMS) is to abstract from the details of: Storage access Storage management Application MDBMS • Location of the MDBMS: Embedded between the application domain and the device domain Device System • Integration into the system: Through operating system Communication components Application MDBMS Operating system Chapter 4.3: Multimedia Database Systems Communication components Page 3 Lehrstuhl für Informatik 4 Kommunikation und verteilte Systeme Multimedia Database Management System • Persistence of data: Data outlive processing programs and technologies, e.g. companies have to keep data in databases for several decades • Consistent view of data: Synchronization protocols provide a consistent view of data in a multi-user system • Security of data: Transaction concepts ensure security and integrity protection in case of system failure. Recovery of lost data. • Query and retrieval of data: Query languages such as SQL (Structured Query Language) enable formulating database queries Each entry has its state information that can be retrieved correctly Chapter 4.3: Multimedia Database Systems Page 4 Lehrstuhl für Informatik 4 Kommunikation und verteilte Systeme Characteristics of MDBMS • Corresponding storage media Multimedia data must be stored and managed according to the specific characteristics of the available storage media • Descriptive search methods Query of multimedia data should base on a descriptive and content-oriented search, e.g. “Picture of a woman with a red scarf” • Device-independent interface Hide details of device control, but offer information on specific characteristics of available storage media (read-only, write-once, write-many) • Format-independent interface DBMS must hide internal storage format and offer conversions to formats requested by the applications (GIF, TIFF, JPEG, ....) This allows changing to new storage technologies without any impact on multimedia applications Chapter 4.3: Multimedia Database Systems Page 5 Lehrstuhl für Informatik 4 Kommunikation und verteilte Systeme Characteristics of MDBMS • View-specific and simultaneous data access Allows consistent, multiple and simultaneous data access through different queries of several applications (e.g. shared editing) • Management of large amounts of data DBMS must be capable of handling and managing large amounts of data. Need of appropriate referencing mechanisms. • Relational consistency of data management Relations among data of one or different media must stay consistent corresponding to their specification. MMDBMS manages the following relations: • Attribute Relation: supports different presentation (audio, video, image) of one object • Component Relation: includes all parts belonging to one data object • Substitution Relation: defines different kinds of presentation of the same information, e.g. equation as tables, graphs, animation • Synchronization Relation: describes temporal relations between data units, e.g. lip synchronization of audio and video Chapter 4.3: Multimedia Database Systems Page 6 Lehrstuhl für Informatik 4 Kommunikation und verteilte Systeme Characteristics of MDBMS • Real-time data transfer DBMS must perform read and write operations of continuous data in real-time The data transfer of continuous data has a higher priority than other database management actions Primitives of multimedia operating systems should be used to support the realtime transfer of continuous data • Long transactions The transfer of large amounts of data will take a long time and must be done in a reliable fashion Chapter 4.3: Multimedia Database Systems Page 7 Lehrstuhl für Informatik 4 Kommunikation und verteilte Systeme Characteristics of MDBMS • Relation between the operating system and MDBMS: The operating system provides the management interface for MDBMS to all local devices The MDBMS provides an abstraction of the stored data and their equivalent devices, as is the case in DBMS without multimedia The communication system provides for MDBMS abstractions for communication with entities at remote computers Operating system and communication system can unify all the different abstractions and offer them Chapter 4.3: Multimedia Database Systems Page 8 Lehrstuhl für Informatik 4 Kommunikation und verteilte Systeme Data Structure Data can be stored in databases as • Unstructured (unformatted): data are presented in a unit where content cannot be retrieved by accessing any structural detail Example: “Mr. Penguin is a student in the seventh term.” • Structured form (formatted): data are stored in variables, field or attributes with corresponding values Example: o.student.surname = “mustername” o.student.name = “hermann” o.student.age = 41 Chapter 4.3: Multimedia Database Systems Page 9 Lehrstuhl für Informatik 4 Kommunikation und verteilte Systeme Data Types Multimedia data can be stored in databases as raw, registering and descriptive data types: • Raw Data: represent the unformatted information content, e.g. letters, pixel, values • Registering data: necessary for correct interpretation and identification of the data; usually concealed in the header. For example: format-description (GIF, TIFF, JPEG, ASCII, EBCDIC, ...), compressed/uncompressed data, etc. • Descriptive data: information about content and structure of the multimedia data to make use easier and faster, e.g. semantic search Chapter 4.3: Multimedia Database Systems Page 10 Lehrstuhl für Informatik 4 Kommunikation und verteilte Systeme Data Types - Examples Text • Characters represent raw data • Registering data describe the coding (e.g., ASCII) • Descriptive data may include information for layout and logical structuring of the text, or keywords Image • Pixels represent raw data • Registering data include the height and width of the picture • Descriptive data are individual lines, surfaces and subjects Chapter 4.3: Multimedia Database Systems Page 11 Lehrstuhl für Informatik 4 Kommunikation und verteilte Systeme Data Types - Examples Video sequence • Pixel matrices represent the raw data • Registering data provides, in addition to other information, the number of images per second • Descriptive data provide a scene description, e.g. ”Jan‘s birthday party” Audio sequence • The digital sample values created by a simple PCM coding represent the raw data • Registering data represent the properties of the audio coding • Descriptive data represent the content of the audio Chapter 4.3: Multimedia Database Systems Page 12 Lehrstuhl für Informatik 4 Kommunikation und verteilte Systeme Operations on Data An MDBMS must offer, for all data types, corresponding operations for: • archival and • retrieval The media related operations will be handled as part or extension of query languages, e.g. SQL Different classes of operations are needed: • input • output • modification • deletion • comparison • evaluation Chapter 4.3: Multimedia Database Systems Page 13 Lehrstuhl für Informatik 4 Kommunikation und verteilte Systeme Operations on Data • Input (insert / record) operation: Data will be written to the database The raw and registering data are always needed, descriptive data can be attached later • Output (play) operation: Read the raw data from the database according to the registering data • Modification: Changing of raw, registering and descriptive data Modification can also be understood as a data conversion from one format to another • Deletion operation: Remove an entry from the database The consistency of the data must be preserved Chapter 4.3: Multimedia Database Systems Page 14 Lehrstuhl für Informatik 4 Kommunikation und verteilte Systeme Operations on Data • Comparison: Many queries to the MDBMS consist of a search and retrieval of the stored data Queries are based on comparison information Individual patterns in the particular medium are compared with the stored raw data → not successful enough → Pattern matching, search in descriptive data, etc. • Evaluation: Generation of the corresponding descriptive data from the raw and registering data Chapter 4.3: Multimedia Database Systems Page 15 Lehrstuhl für Informatik 4 Kommunikation und verteilte Systeme Integration in a Database Model Design of multimedia database systems base on two different kinds of DBMS: • ERDBMS (Extensible Relational Database Management System): Definition of additional, application-dependent data types as domains for attributes Definition of new functions to control behavior of and access to the data Embedding new types and functions into existing RDMBS • OODBMS (Object-Oriented Database Management System): Different media are represented by classes, whose instance variables include the data as internal state Class hierarchy allows object relations, offers well information navigation and flexible presentation possibilities Chapter 4.3: Multimedia Database Systems Page 16 Lehrstuhl für Informatik 4 Kommunikation und verteilte Systeme Relational Database Model Simplest possibility to implement a multimedia database is to use the relational database model • The attributes of different media in relational databases are defined • Attributes can specify text audio video • Advantage compatibility with existent database applications Chapter 4.3: Multimedia Database Systems Page 17 Lehrstuhl für Informatik 4 Kommunikation und verteilte Systeme Relational Database Model - Example A relation “student” is given: Student ( Admission_Number Name Picture Exercise_Device_1 Exercise_Device_2 ) Integer, String, Image, Video, Video • A relation’s attributes can be specified through different media types: image, exercise, video • Other entries are ”athletics”, ”swimming” and ”analysis” Chapter 4.3: Multimedia Database Systems Athletics ( Admission_Number Qualification The_High_Jump The_Mile_Run ) Integer, Integer, Video, Video Swimming( Admission_Number Crawl ) Integer, Video Analysis ( Qualification Error_Pattern Comment ) Integer, String, Audio Page 18 Lehrstuhl für Informatik 4 Kommunikation und verteilte Systeme Relational Database Model • Type 1 Relational Model Value of a certain attribute can be fixed over the particular set of the corresponding attribute types, e.g. the frame rate of the video can be fixed In the example, the videos from the exercise devices 1 and 2 will play at the fixed rate defined by the type 1 specification • Type 2 Relational Model A variable number of entries can be defined through the type 2 relational model In the example, the individual disciplines of each admitted student are identified through their admission numbers • Type 3 Relational Model Additionally, an entry can simultaneously belong to several relations In the example, a video entry of a student can be assigned to the relation ”athletics” as well as to the relation ”analysis” Chapter 4.3: Multimedia Database Systems Page 19 Lehrstuhl für Informatik 4 Kommunikation und verteilte Systeme Object-oriented Database Model In object-oriented databases... • classes with objects are defined • objects can be put in relations via a class hierarchy • a semantic specialization of classes and objects can follow Example • Main class: sports institute • Subclass: athletics, swimming • Objects: students Advantage: • These system offer good information navigation and flexible presentation possibility sports institute athletics swimming students Disadvantage • Query operations are incompletely supported Chapter 4.3: Multimedia Database Systems Page 20 Lehrstuhl für Informatik 4 Kommunikation und verteilte Systeme Conclusions Media can be stored in many ways, only a few were talked about:: Optical Storage • Compact Disc (CD) for multimedia data like images • Digital Versatile Disc (DVD) for videos as main medium • Important: new error protection mechanisms like Reed-Solomon code Hard disk / file systems • Different requirements to between “normal” data and multimedia data • Suitable data structuring needed • New strategies for disk scheduling needed Databases • Can base on relational or object-oriented models • Structure and behavior of data has to be stored • New language primitives are needed • Needed here: more work on transaction management and content-based retrieval of data Chapter 4.3: Multimedia Database Systems Page 21